Trajectory Data-Driven Stop Point Identification of Last-Mile Delivery Electric Tricycle

The explosive growth of express business brings great challenges to last-mile delivery. Designing delivery routes to meet dynamic demands can efficiently integrate delivery resources. Identifying and extracting the characteristics of last-mile delivery stop points is the basis for analyzing the dynamic demands. We take the electric tricycle trajectory points of the Shunfeng Express network in Suzhou as an example and propose three methods, namely, POI screening method, DBSCAN, and DBSCAN-SVM algorithm, to identify the delivery stop points, respectively. By comparing these three methods in terms of accuracy, misidentified points, incorrectly eliminated real points, and algorithm running time, we find that the DBSCAN-SVM is the best in application.